Professional Documents
Culture Documents
Evolution of Busbar Design For Aluminium Reduction Cells PDF
Evolution of Busbar Design For Aluminium Reduction Cells PDF
Equal distribution of current into each of the cathode blocks (collector bars) and
Equal distribution of current to each end (upstream & downstream) of each
collector bar.
These criteria are achieved by providing equi-resistive busbar paths, where the degrees of
freedom are the cross-sectional area of the various busbars, the path length, and (rarely) the
resistivity of the busbar material itself*. The cross-sectional area of the aluminium busbar is
also constrained by resistive heating, which may limit the safe working temperature to
around 2000C and the current density to a maximum of around 100A/cm2.
An Efficient Means of Isolation
As all cells in the potline are in series connection, shutdown of any cell for cathode relining
requires that it be isolated from the circuit via an efficient means of electrical bypass. The
current by-pass is an integral part of the busbar design, and has tended to become more
complex as cells have increased in size / current, and particularly by the implementation of
side risers as opposed to simple end riser configurations.
Safety in Operation
Modern potlines are operating at DC currents to 350 kiloamperes and voltages to 1500
volts, and they have considerable stored energy. As operators and machinery are frequently
in contact with the busbars, their isolation from potential earths is of paramount
importance to achieve a safe working environment.
Minimum Capital Cost
Typically a bus bar system for a modern smelter is made of aluminium, weighs 15,000t and
costs $50m. As such it represents 10-15% of the potline cost. With increasing amperage
the bus bar complexity must also increase in order to avoid high magnetic field gradients.
Economic busbar design is inevitably a compromise between the mass of busbar required in
order to achieve optimum electrical and magnetic field balances, against the minimum
required to achieve acceptable cell performance.
The conceptual design (as developed via modeling) will consider the interaction of key cost
drivers and their impact on cell performance, such as:
The optimum number of anode risers
Routing of the upstream cathode current either around or under the cell
The spacing between the cells
The average and maximum busbar current density consistent with the busbar
rating and the required electrical balances in the network.
The detailed design will further consider fabrication issues, weld design etc in order to
optimise the capital cost. In addition, the voltage drop within the bus bar system must be
considered as a trade-off between the initial capital cost and an on-going operating cost.
Sound design and integrity of electrical joints is particularly important.
Different collector bar designs or connections could also be used as a means to achieve uniformity of
cathode current, but are yet to be actively pursued to the authors knowledge.
4/2
4/2
94-96
4.1-4.3
13.2
93-95
4.2-4.3
13.5
91-94
4.4-4.6
14.5
88-93
4.5-4.7
15.0
Current Efficiency
Volts per Pot
Energy consumption DCkWh/kg
2/8
0.72 (Chinese)
0.80 (Russian)
0.85 (Western)
0.80-0.85
0.80-0.88
0.70-0.80
2/5
Mostly yes
Mostly no
No
Mostly yes
180 - 220
30 - 50
15 - 40
6 / 20
AP30/35
Alcoa A817
Rusal VAMI C255/C280
GAMI GP320
SAMI SY300
Hydro HAL250
AP18/21
Alcoa 697
Comalco-Dubal CD20
VAW CA180
Reynolds P20
Kaiser P80
Kaiser P69
Reynolds P19
Sumitomo
Nippon
Mitsui
Alusuisse
Alcan
Hydro
VAW
VAMI
Cell Technologies
150 - 200
20 - 40
15 - 30
5 / 15
250-350
180-220
150-180
120-180
Current Range kA
100 - 150
80 - 120
120 - 180
10 / 25
Side (4, 5 or 6)
Side (2 or 4)
End (2 or 4)
Riser Layout
100 - 150
100 - 200
80 - 120
10 / 25
Side to side
Side to side
Side to side
End to end
Cell Layout
1990-today
1980-today
1970-1985
1965-1980
When Installed
The busbar design capability is not complete however without the support of a reliable
MHD stability prediction. With modern computational power, very complex stability
models are now possible, accounting for the influences of all the cell bus bar system,
neighboring lines, shell shielding, anode consumption and interface deformation. Two main
families of models are most commonly used. The first uses shallow water theory to
analyse wave propagation at the bath-metal interface8,9, and calculates wave growth rates
(doubling times) for the most unstable waveforms to compare different designs. The second
approach is to develop full 3-D models that treat in detail the geometry and generally treat
the background flow with a multiphase three-dimensional model10,11. In these models an
artificial perturbation is induced on the metal surface, and the damping response monitored
via fluctuation in the anode currents (ACD).
Whereas the Pechiney AP30 technology12 has dominated greenfield smelter developments
for more than a decade, there are now at least five commercial technologies operating in the
+300 kA current range. The Russian and Chinese technologies are being used in national
projects, and the latter are also being actively marketed for external application. Each of
these technologies has different busbar configurations, and all designs are underpinned by
MHD modelling.
The modern approach in design of the magnetic fields is to:
Precisely account for the attenuation of the magnetic field by the shell and cradles.
Model predictions are validated against in-situ measurement of the magnetic field
within the cavity of operating cells.
Avoid high Bz gradients in the central part of the cell, which might be generated by
the location and proximity of under-cell busbars or riser flexibles for example.
Study the effect of specific anode changes on cell stability, and optimise the
magnetic field distribution accordingly.
These objectives are achieved by the discretionary spatial location of busbars (x, y and z
coordinates) and the current flow within them, within the constraints imposed by
economics and the need to achieve current balance. Some aspects to consider for side-toside cells are that
The impact of magnetic fields is reduced when the busbars are located further from
the metal pool for example, considering the effect of horizontal conductors (riser
flexes, anode bridge, under-cell cathode busbars) on Bz.
Bz fields can be reduced and balanced by current flowing in parallel busbars rather
than a single bar, so that the Bz effect of passing current around the end of the cell
for example can be reduced if some current is passed inboard of the end and below
the shell.
Directional current flows in the cathode busbars between cells (eg considering the
downstream cathode bus of the upstream cell in proximity with the upstream
cathode bus of the downstream cell) may be used to balance fields.
By fields will be minimised by placing the busbars at the elevation of the metal
pool.
Pre-1920
Pre-1970
Cell Layout
Small cells
Side to side
End risers
End to end
End risers
Side to side
End risers
Large cells
Side to side
Side risers
Drivers
Busbar economy
(Cathode collector
bus at end of cell)
Hooding of cells
MHD stability
Feeding from
vehicles
Computer modeling
Pneumatic conveying
of alumina
Enabling
Technologies
1970-1985
1980-
(b) End-to end cells with asymmetric busbars compensating neighbour row
(d) Simple busbar layout for side-toside cells with side risers. Symmetric
busbar, current to ~ 250 kA.
10
The basic design has evolved whereby cathode collector bars are connected to side risers
and the anode bus. Within this basic configuration some options include
a. Equi-resistive paths created between individual cathode bars and anode jacks on a
few anodes, using busbars of differing cross-section. This approach was taken when
the first computer models focused on the importance of electrical balance, and
sometimes extended to individual control of each anode height. This approach has
proven to be expensive and provides little value to cell performance however, as the
main resistance within the circuit from the anode riser to the bath is reasonably high
and circuits are readily equalized without the need for elaborate anode feeder bus.
b. Connection of groups of cathode bars to common busbars, with the upstream
cathode busbar passing around the cell, and the risers connecting to a common
anode bus. The cathode bars are sized by length and / or cross-section in order to
give the required electrical resistance and current flow. This design is typical of end
riser cells and side-riser cells to 200 kA.
c. As above, but with some upstream cathode busbar passing under the cell to
minimise the required path length. This approach has developed as cells have
become larger, to minimise both the distance as well as the additional resistance that
needs to be built into the downstream path to achieve a reasonable current balance.
The spatial location of these under cell bus bars can be used to balance Bz fields
and, if placed asymmetrically, to balance the Bz contribution from the neighbour
row.
In addition there are important variants between the various technologies for:
The number, location and shape of the anode risers
The elevation of the cathode busbars with respect to the shell
The design of the collector bars (single, twin and/or split; square, rectangular or
round cross-section; cast iron, glue or rammed paste sealing etc)
The connection of the steel cathode bars to the busbar flexibles (bolted copper tabs,
permanent transition joints with welded collector bar connections, flash-welding of
aluminium flex to the collector bars etc)
The busbar sections (cast or machined, straight or tapered, single or multiple leaf)
The busbar connections (bended, bolted joints, stacked plate welds, narrow gap
welds, chemical welds etc)
The degree of symmetry in the busbar layout and / or current flow in the cathode
busbars
An efficient electrical balance will achieve cathode currents with a coefficient of variation
of less than 10%* for individual values, and an upstream-downstream current split within
the range 48-52%.
*
COV=(Standard deviation*100)/Mean
11
Preferably, to bypass the cell without the need to take the potline off load. This has
become more important as potlines have increased in current and size, as the
production loss from frequent switching of cells in or out of circuit can be
substantial. The time to implement rectifier tap changes in bringing the potline back
to normal current also increases as the operating current becomes higher. The
impact of temporary shedding of a large quantum of power into the supply grid, or
the time required to manage temporary power curtailment from a captive power
plant, may also be issues depending on the smelter location.
The bypass operation must be safe, not labour intensive, and able to be deployed
rapidly. The bypass will involve the making of temporary joints across busbars,
typically by driving bridging wedges between existing busbars or by making bolted
or clamped connections to them. The connection points must have safe access and
be amenable to monitoring to ensure that electrical resistance and temperature
remain within acceptable limits. Depending on the design, making or breaking of
the bypass can take from 5-15 minutes. If the bypass operation requires the potline
to be taken off-load, the downtime must be minimized to avoid disruption to the
heat balance and stability of other cells.
The bypass design must be cost-effective. In the bypass mode, some existing busbars
may be called upon to carry more than their normal current while others become
redundant. In some cases, additional bars are fitted (either permanently or
temporarily) to carry part of the bypass current. The design will need to consider the
maximum current density that can safely be applied to busbars and connection
joints, while economizing on the mass of busbar employed.
The bypass design must preserve the MHD stability of neighbour cells. The bypass
design will involve a redistribution of current in the busbars and will generally
impact upon the current balance of both the upstream and downstream cells. This
can have a negative effect on the operation of these cells.
A typical redistribution of busbar current when a cell is in bypass mode is indicated in the
modeling Case Study, Figure A3.
12
Safety in Operation
Accordingly, the immediate working floor around the cells is isolated from earth, and any
connections between live cell components and earth, such as fume ducts and supports for
busbars and potshells, are provided with electrical insulation. Pot tending cranes have
several levels of isolation between the mounting rails (earthed) and the hooks, tooling and
operator cabin that may be in contact with live cells and busbars.
Wall claddings are also typically insulated at lower levels near the working floor. The
potroom building frame is at earth potential however, and may be as near as 3 metres from
live busbars. Damage to structural concrete leading to exposure of metal reinforcing is a
particular hazard. The basement floor is also at earth potential, and typically within arms
reach of live busbars. Specific and strict operating procedures are therefore required to
support design safeguards.
Some of the procedures that will be in place in modern potrooms to enhance electrical
safety include:
Earthing of the potline at the end crossover, to effectively halve the total voltage
drop. Continuous monitoring to detect any shift in the null point location will help
identify any transient earths that may appear. Such earths must be found and
removed.
13
Maintenance priority on issues that may create earths, such as cleaning of molten
metal spillages from floors, prevention of rainwater access, use of rapid setting
resins for floor repairs etc
There are space limitations on the downstream side as cell spacing is reduced,
particularly for wrap around busbars.
Besides choosing a spatial configuration there are a number of other features that can
impact on capital cost. These include
Minimising the mass of busbar required, through choice of high current density as
aluminium, or choice of more conductive materials such as copper.
Detailed design of the bus bar to minimise use of expensive expansion joints.
Detailed design of the system to minimise number and type of welds for benefits
in both capital cost and operating cost (by lowering the busbar resistance).
There are space limitations on the downstream side as cell spacing is reduced,
particularly for wrap around busbars.
Structural design of the basement, shell supports and bus bar supports to be
compatible with the required spatial configuration and any likely upgrades, to avoid
designs that might trap any overflow of bath or metal into the basement, and to
avoid any configuration that might comprise safety.
Bus bars are constructed from solid heavy aluminium sections. These are usually
horizontally cast. Flexible joints between the bus bar, the cathodes and the cell anode beam
are usually made from rolled aluminium coil. Usually the joints are welded.
Welding issues include
Welding heavy sections is not simple. Challenges include controlling the weld pool,
distortion and cracking. Techniques for new bus bar systems include electro-slag
14
welding as practiced in Russia and joining with a series of small (stack) plates using
MIG argon shield welding which is the preferred method elsewhere.
Each method is expensive. Typically a large stack plate weld requires 8 to 10 hours
to complete as each stack plate is cut and ground to size and the previous weld is
also ground to achieve the required fit-up for the next plate. In addition the stack
plate system only achieves a 50 to 60% electrical connection to the parent metal.
This results in electrical joint loss.
One approach to reduce the impact is to redesign the busbar to minimize the number of
welds required and to minimize the number of site welds required. Another approach is to
develop a new weld system. One such method, called narrow gap welding, has been
pioneered by the CSIRO Division of Manufacturing together with the CRC for Welding
Structures in Australia for both carbon steel and aluminium, and has application in both
potshells and aluminium bus bar systems. Narrow gap welding uses a computer controlled
MIG argon shield welding head to control and optimize the formation of the weld pool.
The technique has been proven in both shop and field environments in work sponsored by
Comalco for the construction of new cells at Boyne Smelters Ltd14. With the development
of advanced heads and a four axis automated guidance system, the technique is now in
commercial use15,16. Using this technique it is possible to weld with an 18mm gap, with
controlled shrinkage and minimal weld distortion when joining full size busbar sections.
The manhours required is cut by 300%, due to reduced weld preparation requirements
compared to the conventional approach, and a 100% rather than 50-60% electrical joint
connection is achieved.
Welding also has additional challenges when joining in-situ busbar within an electrical
field, as the presence of the electrical field influences the weld metal pool, making it
difficult to achieve a satisfactory weld. Stopping the electrical current for an extended
period is not an economic option and methods of welding insitu have to be used. Some
techniques that have been tried include:
15
More recently the move to double anode changing, using either two separate anodes or two
anodes on the one rod and a specialised Pacman cleaning device has narrowed the
selection of anode riser - anode combinations.
Suitable combinations in commercial cells operating at 280-320 kA include
32 anodes, 4 risers
36 anodes, 4 risers
36 anodes, 4 risers
40 anodes, 5 risers
48 anodes, 6 risers
XXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXX
X X X X X X X X X X X X X X X X X X X X
X X X X X X X X X X X X X X X X X X X X X X XX
16
of the aluminium busbars. If the potline is producing 250,000 tonnes of aluminium at 13.5
ACMWh/t and $20/MWh, the annual cost of the busbar power loss will be around $3.5
million. Given that there is designer discretion of around 20% in the average volumeweighted current density, there is significant leverage in finding the optimum economic
design.
Output from a simple financial model to determine the optimum current density is shown in
Figure 3. It is based on a typical modern potline using 2000 metres of busbar and operating
at a current of 320 kA, with the following inputs:
Busbar installed cost $US 3500 per tonne
Power cost $US 20/MWh
25 year life and discount rate of 10% for calculation of net present cost
With these assumptions, an optimum busbar current density of around 40A/cm2 is
indicated.
Figure 3. Economic Busbar Current Density
Power - Net Present Cost
Busbar Investment
Total Net Present Cost
40
35
30
25
20
15
10
5
0
20
30
40
50
60
70
80
17
done are constrained by the difficulties in welding in the magnetic field, as previously
noted.
Some examples of successful busbar improvements made on an operating potline include
On side-to-side cells with end risers, re-routing some or all of the upstream collector
bus under the cell and outwards along the central axis to avoid high Bz fields at the
upstream corners of the cell20.
There are still many opportunities to increase current in existing potlines by improving the
busbar performance. A typical pathway forward is
18
End risers
Four side risers, with all upstream cathode busbar passing around the cell ends
Four side risers, with some upstream cathode busbar passing under the cell
Figure A1. Schematic Busbar Layouts
The 3-D models were developed by coupling the commercial codes ANSYS and CFX. The
electric-magnetic models were built in ANSYS. Steady state and transient MHD flows
were calculated with CFX. Metal and bath were treated as multiphase flow using the
homogeneous VOF (Volume of Fluid) model to calculate the bath-metal interface. The
studies of cell stability were done using CFX in transient regime.
Figure A2. Model Structures
ANSYS model
CFX model
19
Model Assumptions. Each cell is assumed to operate at 240kA, and a metal level of 200mm.
An ACD of 45mm was assumed for the Side Riser options, whereas a higher ACD was
necessary to obtain stability model convergence for the End Riser option. Each cell uses the
same shell, same anodes and same pot-to-pot spacing of 6.2m. The return line was
considered to be at 60m, resulting in a Bz imbalance of around 8G. For the Side Riser
Under Bus option, this Bz imbalance was compensated by using asymmetric current. Two
pots either side of the target cell were included in the models. The maximum current
density used was 75 A/cm2 for all bars except for anodic busbars and risers where 60 A/cm2
was used due to some bypass situations.
Current Density in Busbars
The current density in the busbar network is shown in Figure A2, while a comparison for
the End Busbar option in operating and bypass mode is shown in Figure A3.
Figure A2. Busbar Current Density
End Riser
Half-model
(A/cm2)
Figure A3. Busbar Current Density in Bypass Mode End Busbar Option
20
Collector
Bars
Std dev
2.05 %
2.97 %
2.44 %
Anode
Rods
Std dev
0.45 %
0.53 %
0.57 %
US
current
49.7 %
49.2 %
49.8 %
DS
current
50.3 %
50.8 %
50.2 %
Average
current density
(A/cm2)
33.2
41.2
39.6
Maximum
current density
(A/cm2)
75.0
75.0
75.0
Magnetic Fields
A comparison of the magnetic fields is shown in Figure A4. Major changes are apparent
when the risers are relocated from the ends to the sides of the cell:
Stronger Bx field along the sides of the cell
Major reduction in By field, a driver of metal velocity and heave over the long
axis of the cell
Major reduction in Bz field at the upstream corners, a driver of MHD instability.
Comparing the two side-riser options, there are subtle but important differences in the Bz
fields. For the option using undercell busbar, there is
A reduction in Bz intensity at the upstream corners
A more antisymmetric field distribution.
Figure A4 Magnetic Fields - Bx By Bz contour maps
End Riser
Bx max = -131.2 G
By max = - 92.1 G
Bz max = +185.8 G
(G)
21
Metal Circulation
Metal flow is reduced in the side riser options as a result of reduction in the strong force
fields associated with current concentration in the end riser design, Figure A5. For the
Under Bus option, the flow is further improved and is also made more symmetric by the Bz
compensation of the neighbour line.
Figure A5. Metal Flow
End Riser
Vmax = 0.267 m/s
Vaverage= 0.108 m/s
DS
US
Side Riser End Bus
Vmax = 0.235 m/s
Vaverage= 0.076 m/s
DS
US
Side Riser Under Bus
Vmax = 0.229 m/s
Vaverage= 0.055 m/s
m/s
m/s
DS
US
m/s
22
Metal Velocities
>2 cm/s & <10 cm/s
46.6 %
63.6 %
75.4 %
Metal Velocities
<2 cm/s
3.9%
7.8%
9.4%
Metal Velocities
>10 cm/s
49.5%
28.6%
15.3%
Metal Heave
The side riser options have a dramatic effect on flattening the metal contour, Figure A6.
This has beneficial consequences for cell operations including current efficiency, metal
purity and gross carbon consumption.
Figure A6. Metal Heave
End Riser
Zmax = 0.262 m
Zmin = 0.103 m
anode shadow = 0.159 m
[m]
Stability
In order to compare the stability of the three designs, anode removal situations were
simulated. The background flow for each technology (typical flow pattern) was first taken
in account by calculating a transient flow analysis prior to the anode removal.
The typical oscillation periods (all anodes present) were calculated by performing a Power
Spectral Density (PSD) analysis of the currents:
Model
End Riser
Side Riser End Bus
Side Riser Under Bus
After the removal of the anode in the highest BZ location for each model (anode 13 for Side
Riser Under Bus model; anode 12 for End Riser and Side Riser End Bus models), the pots
23
were monitored for 150 s each. Resulting transient currents in typical anodes (one in each
pot headwall, one in pot center and one neighboring anode to the one removed) are shown
in Figure A7. Note that the scale in the End Riser model is much wider than in the two Side
Riser models.
Figure A8 shows the oscillating currents only (the average current for each anode is
subtracted from the transient current). Comparing the two Side Riser options, it is seen that
the Under Bus option produces a smaller oscillation and achieves a faster damping of the
waves. Once again, the scale in the End Riser model is much wider than the two Side Riser
models; as may be seen, its behavior regarding instability is much worse than either of the
Side Riser models.
A PSD analysis for the remaining anodes after the removal operation is shown in Figure A9
and in the table below, confirming that the Side Riser Under Bus option has superior
stability
Model
End Riser
Side Riser End Bus
Side Riser Under Bus
During the flow simulation following anode change, the energy transfer was also studied. It
shows a very similar correlation with the spectral analysis. Figure A10 shows the (volume
averaged) energy transfer for each time step, demonstrating the higher level of energy
transfer after the anode removal for the less stable busbar options. The End Riser option
presented a short circuit (metal touching the anodes) just after the anode removal, even at a
higher ACD. This explains the very high value obtained by the integration of PSD and
energy transfer volume for this busbar arrangement, an order of magnitude above the Side
Riser options.
Comparison of busbar mass
Model
End Riser
Side Riser End Bus
Side Riser Under Bus
13344
15451
17030
Anode bus
(kg)
7193
2730
2730
Total
(kg)
Mass/Design
Current
(kg/kA)
24650
102.7
22955
95.6
24534
102.2
End Riser
End Riser
Figure A8. Oscillating Current in Selected Anodes for 150 Seconds After Anode Change
Figure A7. Transient Current in Selected Anodes for 150 Seconds After Anode Change
24
25
PSD Distribution for all models after the anode removal operation
Energy Transfer Volume Averaged before and after the anode removal
26
References
1
R. Huglen, Magnetic Compensation of Alumina Reduction Cells, 11th International Course on Process
Metallurgy of Aluminium, Trondheim, June 1992.
C.W.Hirt, Volume of Fluid (VOF) Method for Free Boundaries, J. Computational Physics 1981
A.D. Sneyd, Interfacial Instabilities in Aluminium Reduction Cells, J. Fluid Mechanics, vol 236 1992
O. Zikanov, Shallow Water Model of Flows in Hall-Hroult Cells, Light Metals 2004
10
11
C.H. Droste, Magnetohydrodynamic Stability Analysis in Reduction Cells, Light Metals 1998.
12
13
V Potocnik & J.W. Evans, Evolution of Busbar Design in Hall-Heroult Cells and its Impact on the Process,
CIM Conference, August 1986
14
G.P. Brookes, Research and Development in a Project Environment, IE Aust Conference Gladstone,
Australia, Sept 2002
15
Austin Engineering Pty Ltd , 173 Cobalt Street, Carole Park, Brisbane, Queensland, Australia. 4300
16
17
H van de Nieuwelaar & M Ashriel, CADWELD Exothermic Welding, Sixth Australasian Aluminium
Smelter Technology Conference, Queenstown New Zealand, November 1998.
18
H. Luechinger, Not all busbars are equal, Aluminium 80, Jan 2004.
19
J Purdie et al, Improving the Stability of the A817 Pot at Portland Aluminium, Seventh Australasian
Aluminium Smelter Technology Conference, Melbourne Australia, November 2001
20
G E da Mota & G J de Andrade, Magnetic Compensation Project at Albras Smelter, Light Metals 2001.
27
21
D Vogelsang, Application of Integrated Simulation Tools for Retrofitting Aluminium Smelters, Fourth
Australasian Aluminium Smelter Technology Conference, Sydney Australia, October 1992
22